Stats Regression paneloptions

Home › Stats-Regression › Panel data

PanelOptions

What a panel fit estimates, and how it reports it.

public sealed record PanelOptions

Properties — WithIntercept prepends a constant column to the regressors; true by default, and it has no counterpart in the reference, where a constant is a column the caller supplies. EntityEffects and TimeEffects are fixed effects' own; false by default. CovarianceType is the covariance of the estimates; Unadjusted by default, the reference's. Debiased counts the coefficients out of the covariance's degrees of freedom and reads the tests against t and F rather than the normal and χ²; true by default, the reference's. ClusterEntity and ClusterTime cluster by the rows' entity and period. Kernel and Bandwidth are Driscoll-Kraay's; Bartlett's and null by default, null choosing ⌊4(T/100)^(2/9)⌋. ConfidenceLevel is the intervals' level, strictly inside (0, 1); 0.95 by default.

Example — Driscoll-Kraay at the default and at a chosen bandwidth.

using Lodestar.Stats.Regression;
using Lodestar.Stats.Regression.Panel;

int[] entities = [1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4];
int[] periods = [2020, 2021, 2022, 2023, 2020, 2021, 2022, 2023, 2020, 2021, 2022, 2023, 2020, 2021, 2022, 2023];
double[] x = [0.5, 1.1, 1.9, 2.4, 1.2, 1.8, 2.9, 3.1, -0.3, 0.4, 0.8, 1.6, 2.0, 2.2, 3.1, 3.9];
double[] y = [2.1, 3.0, 4.2, 4.9, 4.4, 5.1, 6.8, 7.0, 0.2, 1.3, 1.7, 3.1, 6.1, 6.3, 7.9, 9.2];

var design = new PanelDesign(y, x, 1, entities, periods);

var kernel = new PanelOptions { EntityEffects = true, CovarianceType = PanelCovarianceType.Kernel };
PanelSummary automatic = PanelRegression.FixedEffects(design, kernel);
PanelSummary chosen = PanelRegression.FixedEffects(design, kernel with { Bandwidth = 2 });

int? lags = automatic.Bandwidth;                 // => 1
double error = automatic.StandardErrors[1];      // => 0.0228748828…
double wider = chosen.StandardErrors[1];         // => 0.0228760370…

Remarks — the estimators check the options, not the record: a negative bandwidth, an undeclared covariance or kernel, and a confidence level outside (0, 1) are refused by the fit, and so is any option that fit would not read — effects outside fixed effects, cluster settings without the clustered covariance, a kernel or bandwidth without Driscoll-Kraay. A Bartlett or Parzen bandwidth of T or more is refused as the reference refuses it.

Applies to — net10.0, netstandard2.0.

See also — PanelRegression, PanelSummary.